Replace map transforms with pre-owned and reused buffer

[pbarejko]

Description

Perf optimization - replace map transforms with pre-owned and reused buffer

Fixes # (issue)

Type of change

• Bug fix (non-breaking change which fixes an issue)
• New feature (non-breaking change which adds functionality)
• Breaking change (existing functionality will not work without user modification)
• Documentation update

Screenshots

Please attach before and after screenshots of the change if applicable.

Checklist

• I have read and understood the contribution guidelines
• I have run the pre-commit checks with ./isaaclab.sh --format
• I have made corresponding changes to the documentation
• My changes generate no new warnings
• I have added tests that prove my fix is effective or that my feature works
• I have added a changelog fragment under source/<pkg>/changelog.d/ for every touched package (do not edit CHANGELOG.rst or bump extension.toml — CI handles that)
• I have added my name to the CONTRIBUTORS.md or my name already exists there

[greptile-apps]

Greptile Summary

Replaces per-frame mapped OVRTX transform writes with caller-owned Warp buffers and asynchronous GPU-stream-aware writes.

• Adds a persistent object-transform buffer allocated during legacy binding setup and released during shutdown.
• Writes object and camera transforms through OVRTX bindings using DataAccess.ASYNC and the active Warp CUDA stream.
• Adds contract tests covering buffer selection, stream propagation, and cleanup, plus a changelog fragment.

Confidence Score: 5/5

The PR appears safe to merge, with no concrete blocking or non-blocking defects identified in the changed transform-buffer paths.

The persistent object buffer is initialized and released with the legacy renderer lifecycle, while both transform write paths pass the producing Warp stream and rely on the documented blocking OVRTX write contract to preserve buffer validity.

Important Files Changed

Filename	Overview
source/isaaclab_ov/isaaclab_ov/renderers/ovrtx_renderer.py	Introduces persistent object-transform storage and replaces mapped object and camera updates with stream-aware caller-owned writes; no actionable defect was established.
source/isaaclab_ov/test/test_ovrtx_deformable_bindings.py	Adds focused tests confirming the object and camera write paths pass the expected buffers, asynchronous access mode, and CUDA streams.
source/isaaclab_ov/test/test_ovrtx_renderer_contract.py	Extends shutdown coverage to verify the persistent object-transform buffer is released.
source/isaaclab_ov/changelog.d/ovrtx-caller-owned-transform-write.rst	Documents the transform-write optimization accurately.

Sequence Diagram

sequenceDiagram
    participant Newton as Newton state
    participant Warp as Warp CUDA stream
    participant Buffer as Caller-owned transform buffer
    participant OVRTX as OVRTX binding
    Newton->>Warp: Provide body transforms
    Warp->>Buffer: Launch transform kernel
    Buffer->>OVRTX: write(ASYNC, cuda_stream)
    OVRTX->>OVRTX: Wait on producer stream and consume buffer

Loading

_{Reviews (1): Last reviewed commit: "Replace map transforms with pre-owned an..." | Re-trigger Greptile}

[isaaclab-review-bot]

Isaac Lab Review Bot

Reviewed the legacy OVRTX transform update change from mapped attributes to caller-owned GPU buffers passed through binding.write with DataAccess.ASYNC and the active Warp CUDA stream. The object-buffer lifecycle, camera write path, regression tests, and package changelog are consistent with the intended optimization.

• Design and architecture: The persistent object transform buffer remains internal to the legacy renderer path: it is initialized with legacy fields, allocated alongside the object binding, used for transform updates, and released during legacy shutdown. This follows the existing asynchronous OVRTX write pattern used by geometry buffers.
• API: No public API is added, removed, or renamed. The changed fields and methods are private, while the object transform buffer retains the bound object count and wp.mat44d representation used by the prior mapped-write path. The touched package includes a changelog fragment.
• Implementation: The obsolete mapping import and both transform mapping call sites are removed together. Object writes reuse a persistent buffer, camera writes pass the generated buffer directly, and both provide the producing Warp stream to OVRTX. Focused tests verify buffer identity, asynchronous access mode, CUDA stream forwarding, and object-buffer cleanup. The remaining camera allocation is a performance tradeoff rather than a demonstrated correctness problem.

No blocking issues. No inline issue met the actionable-evidence threshold; the assessment above records the review feedback.

Automated review; human maintainers own approval decisions.